Are the fragments that are generated made of up small squares as shown in the slide? If so, why? Wouldn't using triangles offer better accuracy for rendering an image?

@captianFlint, the little square is just a representation. A fragment corresponds to a pixel in the image, and it will be rendered as the pixel after passing some tests such as depth test and alpha blending. so I guess the main reason for fragments to be little squares is that we like to represent image pixels as little squares, which is not a very precise representation ;-)

@Khryl A Pixel is not a little square

;-)

What sort of things can be primitives outside of the examples given (triangles, points, and lines)? Can any polygon be a primitive?

@jaguar, yes, GL_POLYGON can be taken as parameter for gl_Begin(). Also you can consider quadrilaterals as primitives as implied by GL_QUAD.

@jaguar

There are also some really exotic primitives like GL_LINE_LOOP and GL_TRIANGLE_FAN. Here is a nice illustration:

In practice, triangles are used most often, though some subdivision algorithms use quads.

For those that want to learn more about rasterizing primitives:

http://15462.courses.cs.cmu.edu/fall2015/lecture/triangle

http://15462.courses.cs.cmu.edu/fall2015/lecture/pipeline